What is the difference between Adobe RGB and sRGB and related questions.

[janet Smith]

Hi Asher & Nic

I am an experienced photographer but still relatively new to the digital world. Would you mind explaining to me in easy to understand terms the difference between Adobe RGB and sRGB and why it is important. The more I read here, the more I want to learn, the more I am aware I don't know...

Thanks once again for your help.

 

[Asher Kelman]

Hi Jan,

"Adobe RGB" and "sRGB" refer to Red-Green-Blue color spaces, 3D graphs enclosing all the possible colors that can be mapped with image files taken of colored objects.

Each color defined in this 3D graph by the relative contributions of Red, Blue, or Green, brightness and saturation (ie intensity) of the colors that can be represented.

Now some colors that are particularly strong, may not be included within the boundries of the 3D graph because the overal volume of each color space, unlike the unvierse we live in, is constrained.

Adobe RGB is a larger color space. That means more of the colors captured with a digital camera, for example, can be accurately mapped into the 3D Adobe RGB color space. The are other color spaces, such as Prophoto RGB, which have an even larger 3D volume and therefore can contain even more colors.

sRGB, by contrast is a smaller color space. We say that the gamut, or range of colors that can be accurately mapped, is less than that in the Adobe RGB color space or for that matter in the even more generous ProfotoRGB space.

sRGB is used for a lot of drug store and low end commercial print centers as it is less demanding of consumers with low end monitors which are uncalibrated anyway.

You may ask "What happens to the rich colors in Adobe RGB that are outside the boundries of the gamut of sRGB?" The answer is simple but important and part of what is done for all the color-concerned devices we use.

Colors that cannot be accurately mapped because of a device's limited color space, or gamut, are "perceptually" remapped by our color management software (in our computers) so that the picture appears normal.

This process of reassigning the coordinates of a color (RGB values, luminence and color intensity) requires advanced knowledge of color perception and works well in practice. When we look at a pciture on a screen, even with a file in Adobe RGB color space, the picture has to be displayed on the screen which doesn't have the refinement to be able to produce a lot of the colors! So again the "out-of-gamut" colors have to be remapped to what the screen can display.

Even more important is the gamut of the printing device! There is no point in fiddling around in photoshop getting amazing new colors when the printer simply cannot put out that wonderful color.

Here, we want to do the least remapping as possible. So before one goes too far in phortoshop work, one should check the gamut! How, ask in a new post and we'll take that up.

Asher

 

[janet Smith]

Hi Asher

What a wonderfully detailed reply, I will no doubt be back with more questions in due course... in the meantime thank you very much for this.

 

[Asher Kelman]

Hi Janet,

I have just found this very useful article "The Role of working (i.e.color-) Spaces in Adobe Applications" which you can download here . This covers everything you want to know about how Adobe, the producers of Photoshop, Lightroom and more, uses RGB color spaces.

Even if you know all about color paces, you'll find this article infromative.

Asher

 

[janet Smith]

Thank once again Asher, I will read this with interest, and hopefully will understand.....

 

[nicolas claris]

Hi janet

just to illustrate the brilliant Asher's demonstration ;-), think to the Russian dolls and look:

sRGB gamut (red) vs SWOP vs image data (dots)

© www.chromix.com

Some interesting links there too :
http://www.color.org/index.xalter

 

[janet Smith]

Hi Nic

Thanks for this too, so much to learn! No doubt I'll be back with lots of questions, I'm feeling really dim not understanding all this stuff, but I'm sure I'll get my head round it eventually, once again your help is very much appreciated.

 

[KrisCarnmarker]

Janet, I usually send people to the Cambridge in Colour tutorials. I find these extremely well written with informative diagrams and examples. For this particular subject, start with the Colour Management tutorial. At the bottom of the page are links to parts two and three of the tutorial.

 

[janet Smith]

Hi Kris

Many thanks for this, I'm going to have my work cut out this next few days digesting all this....

 

[Paul Bestwick]

Asher,

nice explanation. I am looking to the day when colour management has advanced to the point where cameras capture even more & better. & of course monitors & printers (inks, dyes, papers) are totally in synch with our captures. The way technology moves I don't think we will be waiting too long.

Cheers,

Paul

 

[René Damkot]

Adobe RGB is a larger color space. That means more of the colors captured with a digital camera, for example, can be accurately mapped into the 3D Adobe RGB color space. The are other color spaces, such as Prophoto RGB, which have an even larger 3D volume and therefore can contain even more colors.

One important point to make, is that there are not *more* colors in AdobeRGB then in sRGB, but *different* colors. An 8bpc image will still only have 256 steps per channel to work with. So the amount of colors stays 16.7-and-a-bit million, regardless of colorspace used.

That's why ProPhotoRGB should only be used 16bpc. Too coarse 'steps' otherwise, which cause banding.

I wrote a longish post over on POTN (this one) about color management. Quite a few links in it (at the bottom) explaining (why) different color spaces.

 

[Asher Kelman]

One important point to make, is that there are not *more* colors in AdobeRGB then in sRGB, but *different* colors. An 8bpc image will still only have 256 steps per channel to work with. So the amount of colors stays 16.7-and-a-bit million, regardless of colorspace used.

That's why ProPhotoRGB should only be used 16bpc. Too coarse 'steps' otherwise, which cause banding.

I wrote a longish post over on POTN (this one) about color management. Quite a few links in it (at the bottom) explaining (why) different color spaces.

Yes, René a fine but valid distinction, and your concept is correct too.

Still people get the point. When I say more colors I may like "here are the colors in sRGB , you need a lot of saturated colors in the blue green area, then try this Epson K3 color space, you'll find those colors just in this corner".

"Well we can't actually see them properly on the Eizo monitor, but the Epson K3 printers can do this area well so let's see what we get."

I do not use "more colors" in the sense of the number of colors that one might be able to identify in an 8 BIT file rather the zones on a 3D representation of color space that can be achieved with any device.

Yes they are different uses of that word, but in the context of my writings more is still, to me at least correct! It's like Anglo-Saxon swear words, have a huge breadth and depth of meaning!

I'll now look at your link

Asher

 

[Doug Kerr]

Hi, Asher,

Colors that cannot be accurately mapped because of a device's limited color space, or gamut, are "perceptually" remapped by our color management software . . .

Or according to one of several other strategies, depending on the option we choose when instructing the color management chain (when we have that privilege).

A nice summary of the topic. Thanks.

 

[Andrew Rodney]

Check this out:

http://www.adobe.com/digitalimag/pdfs/phscs2ip_colspace.pdf

Should clear things up.

In nutshell, if you define colors using an 8-bit system, both sRGB and Adobe RGB (all working space and RGB color spaces) define the same numeric system to define colors. But the most saturated color you can create in either is (for example) G255. They are NOT the same colors because the scale isn't the same. A color space gives the same set of numbers a scale within human vision. So, numbers alone cannot provide enough information to define a color. We need number and a color space. The same numbers in different color spaces fall in different locations (scale) of human vision.

 

[Asher Kelman]

Thanks, Andrew,

Yes that's a good article Adobe wrote!

Thanks again for sharing it.

I really wish one could just have and xyz shift to get from one space to another but that would only work approximately in the center and very quickly be screwed up as colors became satuated. A pity!

Asher

 

[Doug Kerr]

What is really unique about Adobe RGB

A basic fact that has (I think) gone unmentioned here is the basic difference between the definition of the Adobe RGB color space and the sRGB color space. This is that, while both spaces involve amost identical defintions of the "red" and "blue" primary chromaticities, the "green" (G) primary of the Adobe RGB space is a different one than that of the sRGB space, farther from the other two primaries (on the CIE chromaticity diagram, the familiar graphic presentation of chromaticity).

Thus the Adobe RGB "chromaticity triangle" on the diagram embraces chromaticities (of the "generally green" flavor) not embraced by the triangle for the sRGB space.

Note that the scope of that triangle, while often thought to give us a presentation of the gamut of a color space, doesn't really. It only tells us the range of chromaticities available if, for each one, we are not beyond some certain luminance. As we increase in the luminance we want, the palette of chromaticity becomes more and more limited, until we reach the highest possible luminance, at which our color can have only one chromaticity: "white" (the reference white for that color space).

For example, we can only have colors whose chromaticity is the same as the green primary of our color space if the luminance we want is not greater than the luminance represented by "100% green, 0% red and blue".

Best regards,

Doug

 

[janet Smith]

Hi Jan,

"Adobe RGB" and "sRGB" refer to Red-Green-Blue color spaces, 3D graphs enclosing all the possible colors that can be mapped with image files taken of colored objects...............Asher

Hi Asher

Are you feeling like Professor Higgins out of "My Fair Lady" because I'm feeling like Eliza Doolittle "By George I think I've got it!!!!!"

I may have to call you Prof from now on ........ I read and understood your brilliant explanation and the article you directed me to, and now understand the concept of Colour Spaces, why they are important and how they need to be applied. I'm sure I still have loads more to learn, but at least I have the basic concept firmly in my grasp now, so thanks to you and to Nic for explaining this and my earlier difficulties with interpolation.

I have 18 months approximately before we relocate to a very remote part of Scotland. It is likely that we may settle on the Isle of Harris in the Outer Hebrides, reputedly the most beautiful of the Scottish islands. So before we disappear into the heather and the mists, I need to learn more, so I'd better get down to some work now.....

Anyway, a very big thank you to all who've helped.

 

[Tim Armes]

They've been some great answers here, however I'm not sure if we've explained why you would use one colour space over the other.

I'll try to simplify.

The sRBG colour space was intended to represent the range of colours that could be reproduces by a typical computer monitor. An image in this colour space should be acceptable when viewed on most screens, and as such is the best colour space to use for web images. Since most people don't have calibrated screens colour reproducation won't be accurate, but you'll be in the right ballpark. Even though many newer screens are capable of displaying more colours than sRGB can encode, this is still a good colour space for its purpose.

AdobeRBG is a wider colour space. It encodes a greater range of colours than sRGB, and as such it's capable of encoding more of the colours that your camera is capable of capturing. It's also able to encode colours that your printer may be able to print that your screen can't show. As such, this is a very good choice of working colour space. The only reason to convert an image from Adobe RGB to sRGB would be to post it on the web.

Note that you should always keep the original in Adobe RBG since the conversion can't be reversed. Since Adobe RGB is wider than sRGB the conversion to sRGB will "crop" the colours to those available in sRGB. There is no way to accurately convert back to the wider colour space since it's not possible to determine the original value of a "cropped" colour.

Some people use ProPhoto RGB, be this should be reserved for those who know what they're doing. The colourspace is incredibly wide, even including non-existant colours. It must be used in 16-bit mode to avoid posterized images.

Tim

 

[Asher Kelman]

Good points Tim! I never said why. Hmm!

One more thing, Janet! There's a very good reason for leaving your files in the very largest color space your camera can deliver. In years to come, printers will be improved and the colors that can be delivered to paper accurately will include saturated colors that we cannot achieve right now.

All the people who have saved images as jpgs have crippled the files in so many dimensions.

So that's it. Use the largest color space you can.

sRGB is just crippling files even though they still give great pictures today. To get all the possible colors you mustn't throw them away today!

Asher

 

[janet Smith]

Hi tim & Asher

Good points, thankfully I have kept all Raw files so at any point I can revert to them. ..

Have a great day.

 

[Tim Armes]

So that's it. Use the largest color space you can.
Asher

I don't completely agree. ProPhoto RGB isn't appropriate for many people. Keeping your RAW images is the best way to ensure that you'll be able to get the most out of your older photographs in the future. However, I'd recommend keeping your working masters in Adobe RGB, or Bruce RGB, and something else that isn't overkill.

Tim

 

[janet Smith]

Hi Tim

Luckily for me I have kept them in Adobe RGB. Again, Thank you.

 

[Andrew Rodney]

Digital cameras don't provide images in color spaces unless you ask it to do so by shooting in JPEG and tossing away the raw. The raw has no color space. The scene gamut does, it might easily (and often does) exceed Adobe RGB (1998) gamut. The raw doesn't care. Unless you toss it.

ProPhoto RGB is absolutely a necessary working space if your goal is to encode images who's scene gamut exceed Adobe RGB (1998) of which there are many. Its no more difficult or dangnerous than Adobe RGB (1998). Its just different. For a reason (expressed above).

If you need to dumb down the system, work in sRGB and be content with the potential color data you don't have any longer.

Note, that if the worlds displays produced Adobe RGB instead of sRGB (something that based on emerging technology could happen some day), sRGB would be a useless and poor color space for this dumbed down workflow and Adobe RGB would not. As mentioned, since so many non calibrated displays roughly behave in some semblance of sRGB, sRGB is the best dumbed down space. But that could and probably will change. Keep in mind that sRGB is based on a circa 1994 CRT display, of which I suspect few are still in use today.

 

[Andrew Rodney]

However, I'd recommend keeping your working masters in Adobe RGB, or Bruce RGB, and something else that isn't overkill.

Tim

Tim, Bruce designed Bruce RGB around 1998 or so, stopped using it a year or so after.

Bruce's working space of choice was ProPhoto RGB and he was greatly responsible for aiding Kodak in its design, use and promotion.

 

[Doug Kerr]

Hi, Andrew,

The raw has no color space

True in a CFA (color filter array) camera, and not surprising, since the raw output of such does not present us with a "color" (that is, non-monochrome) image - just a data set from which we can infer a "full color" image with our choice of demosaicing algorithms. (Note, though that there are of course monochrome color spaces.)

A camera with a bona fide tristimulus output (such as a Sigma SD14) does deliver its raw output in a color space.

Best regards,

Doug

 

[Bart_van_der_Wolf]

A camera with a bona fide tristimulus output (such as a Sigma SD14) does deliver its raw output in a color space.

A bona fide tri-stimulus output is almost a bit of a contradiction in terms, wouldn't you agree

?

Besides, the Sigma implementation of the Foveon sensor technology lacks the filtration needed for a perceptually accurate spectral reproduction.

Bart

 

[Bart_van_der_Wolf]

Keep in mind that sRGB is based on a circa 1994 CRT display, of which I suspect few are still in use today.

While true, the advent of early LCDs wasn't necessarily a step up, gamut wise. My better than average LaCie 321 (admittedly not an ADOBE RGB targeted EIZO display) approximately equals (betters in a few areas) sRGB (as in my previous Trinitron CRT).

Bart

 

[Asher Kelman]

While true, the advent of early LCDs wasn't necessarily a step up, gamut wise. My better than average LaCie 321 (admittedly not an ADOBE RGB targeted EIZO display) approximately equals (betters in a few areas) sRGB (as in my previous Trinitron CRT).

Bart

Worse than that Bart!

Go 2 inches from the center and so on until the side and compare the measurements! The results will dissapoint you. But you know that!

I have already decided that one has to place one's important color object in the center of the screen to be sure of the most accurate color.

One fact of life is that anywhere near the sides, the LCD screens are whacky!

Asher

In truth the old fashioned LaCie monitor stands it's own today. However it still emits some radiation so it's not for me!

 

[Doug Kerr]

Hi, Bart,

A bona fide tri-stimulus output is almost a bit of a contradiction in terms, wouldn't you agree

How's that?

Best regards,

Doug

 

[Andrew Rodney]

While true, the advent of early LCDs wasn't necessarily a step up, gamut wise. My better than average LaCie 321 (admittedly not an ADOBE RGB targeted EIZO display) approximately equals (betters in a few areas) sRGB (as in my previous Trinitron CRT).

Bart

Ah but there's way more to this than the gamut. The so called gamma (tone response curve) of CRTs and LCD's is quite different although the manufacturers have tried to make LCD's appear to behave the same way. The chromaticity of the two technologies are different. The target luminance possible between the two are vastly different.